In this Al Arabiya interview, Dr. Dani Geagea examines a central legal conflict of the generative-AI era: how copyright law should distinguish the source of training material, the act of model training and the content produced for users.
The decisive question may arise before training begins
The dispute cannot be reduced to the abstract question of whether machine learning is transformative. A court may need to examine how each protected work was obtained, whether the relevant copy was authorised, what was retained, how it was used and whether any protected expression later appeared in an output.
One dispute, several legal layers
Was the material lawfully licensed or obtained, or do the publishers prove unlawful copying or distribution?
Fair use is a fact-specific defence under United States law, not an automatic immunity for every form of AI training.
Reproducing protected lyrics or removing copyright-management information may raise questions distinct from the training process itself.
Why the position of independent creators matters
Major publishers can retain specialist lawyers and technical experts. Independent writers and artists may not even know that their work was used. The interview therefore considers whether collective licensing, transparent record-keeping and practical output safeguards could provide a more realistic structure than requiring every creator to litigate separately against every technology company.
This page provides general legal commentary. It does not state that the allegations in the pending proceedings have been proved.